Private Prediction Sets

arXiv:2102.06202v3 Announce Type: replace-cross
Abstract: In real-world settings involving consequential decision-making, the deployment of machine learning systems generally requires both reliable uncertainty quantification and protection of individuals' privacy. We present a framework that treats these two desiderata jointly. Our framework is based on conformal prediction, a methodology that augments predictive models to return prediction sets that provide uncertainty quantification -- they provably cover the true response with a user-specified probability, such as 90%. One might hope that when used with …

arxiv cs.ai cs.cr cs.lg decision deployment framework machine machine learning making methodology prediction privacy private protection quantification real return settings stat.me stat.ml systems uncertainty world